X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~mdw/git/disorder/blobdiff_plain/0a4af69255e6fee5274e645e96eee16b6c67a70c..15091f766f1e8c380d2a7f752c387a2cecaf5bb1:/lib/unidata.h

diff --git a/lib/unidata.h b/lib/unidata.h
index f71eb64..aa8814d 100644
--- a/lib/unidata.h
+++ b/lib/unidata.h
@@ -39,6 +39,8 @@ enum unicode_Grapheme_Break {
   unicode_Grapheme_Break_Extend,
   unicode_Grapheme_Break_L,
   unicode_Grapheme_Break_LF,
+  unicode_Grapheme_Break_LV,
+  unicode_Grapheme_Break_LVT,
   unicode_Grapheme_Break_Other,
   unicode_Grapheme_Break_T,
   unicode_Grapheme_Break_V
@@ -72,15 +74,14 @@ enum unicode_Sentence_Break {
 };
 extern const char *const unicode_Sentence_Break_names[];
 enum unicode_flags {
-  unicode_normalize_before_casefold = 1
+  unicode_normalize_before_casefold = 1,
+  unicode_compatibility_decomposition = 2
 };
 
 struct unidata {
-  const uint32_t *compat;
-  const uint32_t *canon;
+  const uint32_t *decomp;
   const uint32_t *casefold;
-  int16_t upper_offset;
-  int16_t lower_offset;
+  const uint32_t *composed;
   unsigned char ccc;
   char general_category;
   uint8_t flags;
@@ -89,6 +90,13 @@ struct unidata {
   char sentence_break;
 };
 extern const struct unidata *const unidata[];
-#define UNICODE_NCHARS 195200
-#define UNICODE_MODULUS 128
+extern const struct unicode_utf8_row {
+  uint8_t count;
+  uint8_t min2, max2;
+} unicode_utf8_valid[];
+#define UNICODE_NCHARS 1114112
+#define UNICODE_MODULUS 16
+#define UNICODE_BREAK_START 196608
+#define UNICODE_BREAK_END 917504
+#define UNICODE_BREAK_TOP 918016
 #endif